home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.physics.fusion
- Path: sparky!uunet!sun-barr!ames!pacbell.com!tandem!zorch!fusion
- From: blue@nscl01.nscl.msu.edu
- Subject: The Rev. Thomas Bayes and Cold Fusion
- Message-ID: <0095FED9.44D732C0.15230@dancer.nscl.msu.edu>
- Sender: scott@zorch.SF-Bay.ORG (Scott Hazen Mueller)
- Reply-To: blue@nscl01.nscl.msu.edu
- Organization: Sci.physics.fusion/Mail Gateway
- Date: Mon, 7 Sep 1992 02:36:27 GMT
- Lines: 37
-
- I stole that title from something I just ran across on page 9 of the
- Jan '92 issue of Physics Today - a commentary by Phillip W. Anderson
- on the application of Bayesian statistics to "needles in haystacks and
- the fifth force." For "fifth force" one may as well substitute "cold
- fusion" and Prof. Anderson's commentary will remain equally valid. I
- think a few quotes will serve to convey the essence, and I will indicate
- my changes in the wording by CAPS.
-
- "These statistics (Bayesian) are the correct way to do inductive reason-
- ing from necessarily imperfect experimental data . . . The essence of
- it is to clearly identify the possible answers, assign reasonable
- a priori to them and then ask which of the answers have been made more
- likely by the data. It's particularly useful for simple "null" answers."
-
- "Let us take COLD FUSION. If we assume from the outset that there is
- COLD FI
- USION and we need only measure its magnitude, we are assigning
- the bin with zero SURPLUS HEAT an infinitesimal probability to begin
- with. Actually we should assign this bin, which is the null hypothesis
- we want to test, some finite a priori probability - like 1/2 - and
- share out the remaining 1/2 amoung all the other possible VALUES.
- We then ask the question, does a given set of statistical measurements
- increase or decrease this share of the probability? It turns out that
- when one adopts this point of view, it often takes a much larger devia-
- tion of the result from zero to begin to decrease the null hypothesis's
- share."
-
- Having read this my conclusion is that those of you trying to make a
- case for cold fusion should find a good book on Bayesian statistics
- or, at the very least, read what Prof. Anderson has to say in its
- entirety.
-
- Correction: The fourth line of the first quote should read a priori
- probabilities to them...
-
- Dick Blue
- NSCL @ MSU
-